2982
Biochemistry 2003, 42, 2982-2994
Mechanistic Characterization of the HDV Genomic Ribozyme: Classifying the Catalytic and Structural Metal Ion Sites within a Multichannel Reaction Mechanism† Shu-ichi Nakano,‡ Andrea L. Cerrone, and Philip C. Bevilacqua* Department of Chemistry, The PennsylVania State UniVersity, UniVersity Park, PennsylVania 16802 ReceiVed September 7, 2002; ReVised Manuscript ReceiVed January 6, 2003
ABSTRACT: Prior studies of the metal ion dependence of the self-cleavage reaction of the HDV genomic ribozyme led to a mechanistic framework in which the ribozyme can self-cleave by multiple Mg2+ ionindependent and -dependent channels [Nakano et al. (2001) Biochemistry 40, 12022]. In particular, channel 2 involves cleavage in the presence of a structural Mg2+ ion without participation of a catalytic divalent metal ion, while channel 3 involves both structural and catalytic Mg2+ ions. In the present study, experiments were performed to probe the nature of the various divalent ion sites and any specificity for Mg2+. A series of alkaline earth metal ions was tested for the ability to catalyze self-cleavage of the ribozyme under conditions that favor either channel 2 or channel 3. Under conditions that populate primarily channel 3, nearly identical Kds were obtained for Mg2+, Ca2+, Ba2+, and Sr2+, with a slight discrimination against Ca2+. In contrast, under conditions that populate primarily channel 2, tighter binding was observed as ion size decreases. Moreover, [Co(NH3)6]3+ was found to be a strong competitive inhibitor of Mg2+ for channel 3 but not for channel 2. The thermal unfolding of the cleaved ribozyme was also examined, and two transitions were found. Urea-dependent studies gave m-values that allowed the lower temperature transition to be assigned to tertiary structure unfolding. The effects of high concentrations of Na+ on the melting temperature for RNA unfolding and the reaction rate revealed ion binding to the folded RNA, with significant competition of Na+ (Hill coefficient of ≈1.5-1.7) for a structural Mg2+ ion and an unusually high intrinsic affinity of the structural ion for the RNA. Taken together, these data support the existence of two different classes of metal ion sites on the ribozyme: a structural site that is inner sphere with a major electrostatic component and a preference for Mg2+, and a weak catalytic site that is outer sphere with little preference for a particular divalent ion.
Metal ions play a variety of roles in the folding of complex RNA structures and in the catalytic activity of ribozymes. Divalent ions have long been recognized as necessary for the tertiary folding of RNA under physiological ionic strengths. For example, yeast tRNAPhe requires 3-6 Mg2+ ions for proper tertiary folding at low ionic strength (1-3), and pseudoknots typically require delocalized divalent metal ions to fold efficiently (4, 5). One of the primary reasons divalent metal ions are required is to neutralize the close approach of negatively charged phosphates that occurs upon compaction of the RNA molecule. Metal ions can be thermodynamically bound at specific sites, “site bound”, or kinetically labile, “diffuse” or “delocalized” (6). In addition, site-bound metal ions can remain fully hydrated, “outer sphere”, or can become partially or fully dehydrated, “inner sphere” (6). Monovalent ions have also been shown to play important roles in both the secondary and tertiary folding of RNAs. For example, high concentrations of monovalent ions, typically 0.1 M or greater, can induce the proper tertiary †
Supported by NIH Grant GM58709 and a Camille Dreyfus Teacher-Scholar Award and Sloan Fellowship to P.C.B. * To whom correspondence should be addressed. Phone: (814) 8633812. Fax: (814) 863-8403. E-mail:
[email protected]. ‡ Current address: High Technology Research Center, Konan University, 8-9-1 Okamoto, Higashinada-ku, Kobe 658-8501, Japan.
folding of many RNAs (6), and monovalent ions can be site bound, with several examples of dehydrated K+ ions known (7-9). The ribozyme from hepatitis delta virus (HDV)1 occurs in closely related genomic and antigenomic forms (10), and a crystal structure is available for the cleaved form of the genomic ribozyme (11, 12). The ribozyme adopts a complex tertiary structure involving nested pseudoknots and a buried active site. The ribozyme can accelerate the rate of cleavage of a phosphodiester bond between positions -1 and +1 and gives 5′-hydroxyl and 2′,3′-cyclic phosphate termini. The mechanism for self-cleavage provided an example of general acid-base catalysis by a ribozyme (13, 14), and general acid-base catalysis has subsequently been suggested in the mechanisms of the ribosome (15) and the hairpin ribozyme (16). A cytosine residue (C75 in the genomic ribozyme, C76 in the antigenomic ribozyme) near the scissile bond has been shown by kinetic studies to have a pKA near neutrality (13, 1 Abbreviations: AS(-30/-7), antisense oligomer complementary to nucleotides -30 to -7; EDTA, ethylenediaminetetraacetic acid; HEPES, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid; G11C, ribozyme used in these studies with a G to C change at position 11; HDV, hepatitis delta virus; PAGE, polyacrylamide gel electrophoresis; TE, 10 mM Tris and 1 mM EDTA (pH 7.5); TEN250, TE with 250 mM NaCl; Tris, tris(hydroxymethyl)aminomethane; UV, ultraviolet.
10.1021/bi026815x CCC: $25.00 © 2003 American Chemical Society Published on Web 02/22/2003
Characterizing HDV Metal Ions Scheme 1: Proposed Proton Transfers in Bond Cleavage
Biochemistry, Vol. 42, No. 10, 2003 2983 Scheme 2: Three-Channel Mechanism for Ribozyme Cleavage
MATERIALS AND METHODS
14) and has been implicated as the general acid in the reaction, acting to protonate the 5′-bridging oxygen (14, 17, 18) (Scheme 1). NMR studies provided evidence that this pKA is not shifted in the product form of the ribozyme, suggesting that it may shift in the transition state or in a high-energy ground state of a properly folded precursor ribozyme (19). A pentahydrated metal hydroxide has been implicated as the general base in the reaction under near-physiological metal ion conditions to deprotonate the 2′-OH nucleophile (14, 20) (Scheme 1). Additionally, proton inventory studies provided evidence for two proton transfers in the reaction (17). Brønsted studies of exogenous base rescue of a C76deleted antigenomic ribozyme indicated that approximately half of this proton is transferred in the transition state (21); likewise, Brønsted studies of the metal dependence of the reaction for the genomic ribozyme indicated that half of a proton is transferred to the general base in the transition state (20). R- and β-values similar to these have been found for cleavage of RNA by the protein enzyme, RNase A (22). To assess the contributions of metal ions to folding and catalysis in the HDV genomic ribozyme, we examined selfcleavage over a wide range of Mg2+ ion concentrations using a metal-buffered system in the presence of high ionic strength (1 M NaCl) (20). The data could be described with a multichannel kinetic model in which each channel involved a different role for Mg2+ ions. Multichannel models have subsequently been applied to the hammerhead ribozyme (23). Surprisingly, the HDV ribozyme was found to cleave in the absence of participating divalent metal ions (channel 1), indicating that divalent ions are not absolutely necessary for folding or catalysis of the ribozyme (14, 20). Channel 2 of the mechanism involves a structural Mg2+ ion, while channel 3 involves both structural and catalytic Mg2+ ions. Assessment of the contributions of these ions to the rate of cleavage in 1 M ionic strength gave ≈125-fold for the structural ion and only ≈25-fold for the catalytic ion (20). However, these studies did not provide insight into the nature of the divalent metal ion sites or any specificity toward Mg2+. In an effort to characterize the divalent metal ion sites, we have probed the effect of changing the metal ion under conditions that populate primarily channel 2 or channel 3. We used a series of mono-, di-, and trivalent metal ions at various concentrations and measured their effects on catalysis and RNA unfolding thermodynamics. Our data support two different classes of metal ion sites on the HDV ribozyme: an inner-sphere structural site and an outer-sphere catalytic site.
Materials. Unless otherwise noted, the ribozyme and solutions were prepared, and the RNA was renatured as described (20, 24). All experiments used the G11C mutant of the -30/99 ribozyme which leads to fast monophasic folding due to destabilization of the Alt P1 misfold (25). Reactions were promoted with the AS(-30/-7) antisense oligonucleotide, which acts to resolve the Alt 1 misfold (24), and were initiated by addition of divalent ion or a divalent ion/Na+ mixture, as appropriate. Kinetic Methods and Data Fitting. In general, kinetics methods followed those previously described (20). Plots of cleavage rate versus divalent ion concentration were fit to linear or logarithmic versions of the Hill equation (eq 1a,b)
kobs )
log
kmax[M]RH/KdRH 1 + [M]RH/KdRH
kmax - kobs ) -RH log [M] + RH log Kd kobs
(1a)
(1b)
where [M] is the concentration of a given ion, kobs is the observed rate constant, kmax is kobs at saturating ion concentrations, RH is the Hill constant, which is a lower limit for the number of ions taken up by the RNA, and Kd is the apparent dissociation constant for ion binding to the RNA. Data for the three-channel model were fit according to Scheme 2, with three linked equilibria. Binding of the structural and catalytic Mg2+ ions is shown as sequential on the basis of previous studies (20), and the binding of the two Na+ ions is treated as cooperative, although the Hill coefficient reveals a measure of actual cooperativity. kobs is the weighted sum of the rate constants for each channel
kobs ) k1f1 + k2f2 + k3f3
(2)
where the weight for channel i is fi, the fractional occupancy. The fractional occupancy of each channel is defined in terms of the partition function, Q
Q)1+
[Na+]2 2 Kd,Na +
+
[Mg2+] [Mg2+]2 + Kd,str Kd,strKd,cat
(3)
so that 2 [Na+]2/Kd,Na + f1 ) Q
(4a)
2984 Biochemistry, Vol. 42, No. 10, 2003 Scheme 3
[Mg2+]/Kd,str Q
(4b)
[Mg2+]2/Kd,strKd,cat Q
(4c)
f2 ) f3 )
Nakano et al.
Thus, in the absence of Mg2+
kobs )
2 k1[Na+]2/Kd,Na + 2 1 + [Na+]2/Kd,Na +
(5)
and when [Na+] , Kd,Na+ 2 log kobs ≈ 2 log [Na+] + log(k1/Kd,Na +)
(6)
and the slope of the steep portion of a log kobs versus log [Na+] plot should approach 2 when the binding of the sodium ions is cooperative. Likewise, under Mg2+ concentrations wherein channel 2 dominates the reaction (k2 . k1; see Results and Discussion).
kobs )
k2[Mg2+]/Kd,str 1+
[Na+]2 2 Kd,Na +
[Mg2+] + Kd,str
(7)
Therefore, this leads to a Kd,app of 2 Kd,app ) Kd,str(1 + [Na+]2/Kd,Na +)
(8)
which is of the form expected for cooperative binding of two Na+ ions competitively inhibiting binding of one structural Mg2+ ion. The equations from the kinetic model for Scheme 2 were derived with the assumption that the three channels are in rapid equilibrium. There are several observations that support this. (A) Metal ions typically bind rapidly to nucleic acids, and their translation along a nucleic acid has no significant barrier (6). Of course, binding of a metal ion to a final folded state can appear slow if binding either requires or induces a conformational change of the RNA with a significant barrier. However, the experiments herein involving Na+ competition were done in 1 M NaCl such that the ribozyme should be largely prefolded into the correct conformation. (B) Binding constants for the inner- and outer-sphere metal ions are similar to those expected from the literature, as pointed out in the text. (C) Binding constants for the inner-sphere metal ion inferred from cleavage studies are in good agreement with those from melting experiments. Nevertheless, we do not have proof that channels are all in rapid equilibrium. As such, binding constants for metal ions should be viewed as apparent. Thermodynamic Methods and Data Fitting. The RNAs used in UV melting studies were of the self-cleaved G11C ribozyme from +1 to 99. This species was prepared by allowing the ribozyme to react to completion in the presence of AS(-30/-7) and Mg2+ (24). Cleaved RNAs were purified by denaturing 5% PAGE and eluted in TEN250. The RNA was ethanol precipitated, washed with 70% ethanol, and stored in TE at -20 °C.
UV melting profiles were obtained at 260 nm on a Gilford Response II spectrophotometer, using a 0.5 cm cuvette and 0.1-0.5 µM RNA. The RNA was renatured prior to melting as described (24). Representative samples were melted at 0.1 and 0.5 µM, and identical results were obtained, consistent with melting of a monomeric RNA species. Heating and cooling curves were superimposable in both Na+ and Mg2+ (for pH 7.0 melts), consistent with reversibility and thermodynamic equilibrium. The heating rate was approximately 1 °C/min, and the data were smoothed over five points before the first derivative of absorbance was calculated with respect to temperature. First derivatives were normalized by dividing by the absorbance at the lowest temperature after smoothing (26). (Low temperature was chosen since absorbance at high temperature can reflect differences in hydrolysis of the RNA backbone, especially in the presence of divalent ions.) The buffer was 25 mM HEPES (pH 7.0) (or pH 8.0 for urea dependence experiments), and the pH was adjusted at room temperature. At pH 8.0, only forward melts were used due to hydrolysis at high temperatures. Melting of the cleaved ribozyme revealed two unfolding transitions which were consistent with unfolding of tertiary and secondary structure (see Results and Discussion). In an effort to obtain thermodynamic parameters for the transitions, the data were fit to linked equilibria for a sequential unfolding model (Scheme 3) involving native (N), intermediate (I), and unfolded (U) species; these appear to correspond to tertiary, secondary, and primary structure, respectively. In general, the data were fit according to the methods of Draper, Giedroc, and co-workers (5, 26, 27). Scheme 3 can be described with a partition function of
Q ) 1 + K1 + K1K2
(9)
and the absorbance can be defined as the sum of ∆At and B
∆At )
(
∆A1K1 (∆A1 + ∆A2)K1K2 + Q Q
)(
(10)
)
BU - BN ∆H1K1 + (∆H1 + ∆H2)K1K2 ∆H1 + ∆H2 Q (11) ∆At is the total hyperchromicity, and ∆A1 and ∆A2 are the fractional hyperchromicities for transitions 1 and 2. B is a baseline function, and BU and BN are the sloping lines for the upper and lower baselines, respectively, with the intermediate treated with an enthalpy-weighted baseline. The slopes and intercepts of the upper and lower baselines were determined independently by linear least-squares fits of the low- and high-temperature portions of the data and were held constant during the actual fit. The baselines were chosen to minimize χ2 of the fit. Thus, only six parameters were adjusted in the fit, namely, ∆H°, TM, and ∆A for the first and second transitions. ∆H°i and TM,i for the ith transition were determined from the van’t Hoff relationship ∆Hi 1 1 (12) Ki ) exp R TM,i T B ) BN +
[ (
)]
Characterizing HDV Metal Ions
Biochemistry, Vol. 42, No. 10, 2003 2985
where R is the ideal gas constant and T is absolute temperature. Curve fitting of absorbance versus temperature data was to the sum of eqs 10 and 11, with Q, BU, BN, K1, and K2 defined parametrically using Kaleidagraph v3.5 (Synergy Software). ∆Si was obtained from TM,i ) ∆Hi/∆Si, and ∆G37,i was from ∆G37,i ) ∆Hi - 310.15∆Si. The m-value of the ith transition, mi, was calculated as the slope of a ∆G37,i(urea) versus urea concentration plot on the basis of the relationship
∆G37,i(urea) ) ∆G37,i + mi[urea]
(13)
where ∆G37,i(urea) is the observed free energy for transition i at a given urea concentration and mi is a thermodynamic parameter that describes the dependence of free energy on urea concentration (28). The dependence of TM on metal ion concentration was analyzed according to the formalism previously developed (5, 27, 29). Briefly, a set of coupled equilibria was used in which all possible ligand-bound forms of the folded (F) and unfolded (U) RNAs involved in a two-state transition are considered. (Note that we consider the simple two-state F T U transition and later extend this to Scheme 3.) The observed equilibrium constant for unfolding (Kobs) is then
ln Kobs ) ln Ko + ln Σu - ln Σf
(14)
where Ko is the observed equilibrium constant in the absence of the ligand and Σu and Σf are binding polynomials for the unfolded and folded states, respectively. A useful form of this equation can be reached by applying the van’t Hoff equation in the absence of ligand, using T0, the TM in the absence of the ligand, and TL, the TM at ligand concentration L
1/TL ) 1/T0 - R/∆H0 ln(Σf/Σu)
(15)
where ∆H0 is the unfolding enthalpy in the absence of ligand. As described by Laing and co-workers (27), the functional forms of Σu and Σf depend on whether the metal ions bind to F only or to both F and U. If the ions bind nonspecifically to F and U, then plots of 1/TL versus ln [L] are expected to be sigmoidal; however, if binding is specific to F, such plots are expected to decrease indefinitely with metal ion concentration at high metal ion concentrations and have a slope of -nFR/∆H0, where nF is the number of ions bound to F (27).
d(1/TL) d(ln [L])
( )
) -nF
R ∆H0
(16)
For this equation, the enthalpies for ion binding to the folded and unfolded states, ∆Hf and ∆Hu, are assumed to be zero. As described, this assumption can introduce small errors in some instances (27). The case of ion binding to F only appears to apply to our data. Fortunately, under these conditions, Σu can be considered to be unity. For specific binding of Mg2+ to independent sites on F
Σf ) (1 + Kf[Mg2+])nF
(17)
where Kf is the association constant for Mg2+ binding to the
native RNA. Note that this model does not take into account any cooperativity (positive or negative) between binding of multiple ions. This formalism was developed for the two-state transition of F T U. For a three-state system (Scheme 3), these equations can still be applied to one of the two transitions if that transition is well isolated from the other or if the data are treated with coupled equilibria. For our purposes, 1/TL for the first transition in Scheme 3 was plotted versus metal ion concentration and fit to eq 15 with ΣI ) 1 and ΣN as defined in eq 17. It should be noted that for the urea dependence experiments an upper baseline could be reasonably well defined, allowing thermodynamic parameters for both transitions to be obtained. However, with increasing Na+ and Mg2+ concentrations in the absence of urea, the extraordinary stability of the ribozyme precluded a good upper baseline from being obtained. In these cases, TL of the first transition was estimated from the maximum of the first derivative; this appears reasonable since the maxima of the two transitions were well-resolved even at high ligand concentration. It is worth pointing out the connection between Schemes 2 and 3. N in Scheme 3 refers collectively to all four states involving R in Scheme 2, and the states involving I and U are not shown explicitly since the kinetic experiments were at temperatures well below the TM for N. However, as pointed out in the text, states involving I may be important in the kinetic mechanism, especially in monovalent ions (20), since the precursor ribozyme may have more unfavorable interactions than the cleaved ribozyme. Implications for this possibility are considered in the text. BACKGROUND Analysis of the binding of metal ions to nucleic acids is complicated by the wide variety of ion sites and binding modes possible, some of which obey mass action and some of which do not (e.g., diffuse ions). Moreover, if the specific ion-free state of the RNA does not have tertiary structure, the contributions of ions to site binding and to folding are typically coupled, leading to the highly cooperative binding of multiple ions (6). This scenario makes classification of the mode of binding of a few structural or catalytic ions difficult, if not impossible. In contrast, binding of ions of RNAs with preformed tertiary structure at high monovalent salt concentrations is typically noncooperative, allowing the opportunity to characterize the mode of ion binding (6). Thus, it is imperative to identify solution conditions under which an RNA molecule is largely folded and the filling of only a few metal ion sites occurs. Experiments for the genomic HDV ribozyme conducted with metal-buffered solutions in 1 M NaCl (20) suggested conditions that might be tried for examining the structural and catalytic metal ions separately. Conditions were sought for which small Hill coefficients occur for filling the structural or catalytic site. For probing the structural site, this condition was fulfilled by reaction conditions under which channel 2 dominates and the catalytic ion is not involved. In particular, conditions of 1 M NaCl and pH 7.0 were found to give divalent ion binding isotherms well described with an RH of 1, suggesting binding of one structural ion (20). In contrast, raising the pH to 8.0 in 1 M
2986 Biochemistry, Vol. 42, No. 10, 2003 NaCl increased the affinity of the catalytic ion and decreased the affinity of the structural ion such that distinctly cooperative binding behavior was observed with an RH of 1.85 (20). This was interpreted in terms of a model in which a structural Mg2+ ion is ≈125-fold more effective than NaCl (even at a concentration of 1 M) at stabilizing the functional tertiary structure, and the catalytic ion binds only to the folded tertiary structure (see sequential ion binding model in Scheme 2) (20). Of practical relevance to the studies here, conditions of 1 M NaCl and pH 7.0 seemed optimal for testing the nature of the structural metal ion. Results with various divalent metal ions and Co(NH3)63+ support this notion (see below). For probing the catalytic site, the RH of unity condition was fulfilled by keeping the structural site saturated at all metal ion concentrations tested. Since the structural site is competitively inhibited by Na+ (see below), this was achieved with a background of low ionic strength. At low ionic strength, an additional concern becomes the occurrence of bulk electrostatic effects in which multiple ions bind simultaneously to induce correct folding of an RNA (6). For tRNA, which is similar in size and buried surface area to the HDV ribozyme (9), these ions bind with Kds of 100170 µM under ionic strengths similar to those used in Figure 1A (6). The Kds observed herein are 10-50 times weaker than this value (Figure 1A, Table 1), suggesting that both specific and nonspecific ions are largely associated with the RNA before the onset of binding in Figure 1A. Moreover, previous studies under these conditions provided evidence that binding of one Mg2+ ion and one H+ ion to C75 of the ribozyme is negatively linked and that this Mg2+ ion inverts the pH profile of the reaction (14, 20). Collectively, these data suggest that rate profiles obtained under low ionic strength conditions are not dominated by bulk electrostatic effects and that these conditions probe the catalytic metal ion. RESULTS AND DISCUSSION Metal Ion Dependence Supports an Outer-Sphere Site for the Catalytic Ion. As described in the Background section, we probed the catalytic ion under conditions of low ionic strength. The first experiments examined the effect of changing Mg2+ to other group IIA metal ions. As shown in Figure 1A, Mg2+, Ca2+, Sr2+, and Ba2+ gave binding isotherms that were well described with an RH of 1, resulting in similar Kd values of 2.4, 4.8, 2.7, and 1.2 mM, respectively (Table 1). Binding was weakest for Ca2+ and tightest for Ba2+ but does not show a monotonic dependence on ionic radius or a large dynamic range in Kd values (Figure 1C) (compare to structural site below). Prior studies from Nishikawa and co-workers using single time points provided qualitative support for similar Kds for Ca2+, Mg2+, and Sr2+ under similar solution conditions (30). The kmax values for Mg2+, Ca2+, Sr2+, and Ba2+ were 3.9, 8.7, 1.5, and 0.069 min-1, respectively (Table 1, Figure 1A). The somewhat larger value of kmax for Ca2+ than Mg2+ is consistent with previous studies (14, 31), although its origin is not understood. Previous studies showed that binding of the structural and catalytic ions is ordered, presumably because binding of the structural ion helps to create the site for the catalytic ion (20). Thus, the drop in rate for Sr2+ and
Nakano et al. Ba2+ may be because the larger ions, even at saturating concentrations, are unable to induce proper local folding of the ribozyme at the structural site. In addition, Ba2+ and Sr2+ have the highest pKAs for their aqua ions (32), making them especially poor general bases in the classical sense, although kmax does not display a monotonic relationship with the pKA of the aqua ion [13.82, 13.18, 12.70, and 11.42 for Ba2+, Sr2+, Ca2+, and Mg2+, respectively (32)], suggesting any underlying metal ion pKA effect may be hidden. To test the catalytic site further, we examined the effect of adding transition metal ions to the reaction. In the presence of 10 mM MnCl2 or CoCl2, the rate was appreciable at 0.77 and 0.34 min-1, respectively (Table 2). Zn2+ ions were also able to support slow reactivity (0.011 min-1). The effects of Mn2+, Co2+, and Zn2+ on the reaction were qualitatively consistent with previous reports (30). In contrast, 10 mM Ni2+ and Cu2+ did not lead to detectable reaction. This may be because these ions tend to interact directly with the bases and may misfold or unfold the ribozyme (33). Reactions were also carried out in the presence of 1 mM Co(NH3)63+ and no added divalent ion and revealed no detectable reaction under similar time courses (Table 2). To characterize Co(NH3)63+ binding further, the binding affinity of Mg2+ was measured in the presence of increasing amounts of Co(NH3)63+ (0, 0.1, 1.0, and 10 mM) (Figure 2A, Table 3). Co(NH3)63+ and Mg(H2O)62+ have similar size and geometry, and Co(NH3)63+ is exchange inert, making Co(NH3)63+ a good mimic of an outer-sphere metal ion (34, 35). It can be seen that the same kmax is reached or approached in the presence of Co(NH3)63+ but that Co(NH3)63+ increases the concentration of Mg2+ required to reach rate saturation. This observation is consistent with previous reports of competitive inhibition by Co(NH3)63+ and supports an outer-sphere or weakly dehydrating site for the catalytic ion (14). Presumably, Co(NH3)63+ is unable to catalyze the reaction because it does not ionize appreciably at pH 7. The data in this section suggest that this ion is not binding into a region of high charge density. In particular, if an innersphere interaction with one or more phosphates was occurring with significant ion desolvation, then one would expect affinity to decrease with ionic radius, as found for the structural ion (see below) (36, 37). Also, it is expected that the Kd for the ion would be significantly smaller (in the micromolar range), especially at low ionic strength, instead of the millimolar range observed here (27). In addition, it is expected that the binding affinity would be sensitive to ionic strength, whereas little or no sensitivity has been found (20). Thus, our data support this ion either remaining fully hydrated as an outer-sphere or diffuse metal ion or undergoing dehydration but binding into a region of low charge density (4). This latter possibility would favor the larger ions, which have smaller enthalpic penalties for dehydration (32), as well as provide weak discrimination against Ca2+ since Ca2+ has a hydration number of 8 and Mg2+, Sr2+, and Ba2+ have hydration numbers of 6 (32). However, competitive inhibition by Co(NH3)63+ suggests that any dehydration is not critical for binding. Crystallographic studies on the cleaved form of the ribozyme did not reveal a metal ion with high occupancy at the cleavage site, although weak electron density was seen nearby (11, 12). This could arise in part because metal ion
Characterizing HDV Metal Ions
Biochemistry, Vol. 42, No. 10, 2003 2987 Table 1: Constants for Divalent Cation Binding to the Ribozyme at pH 7.0 RH
M2+ 0 M NaCl Mg2+ Ca2+ Sr2+ Ba2+ 1 M NaCl Mg2+ Ca2+ Sr2+ Ba2+
Kd,Hill (mM)
kmax (min-1)
2.4 ( 0.4 4.8 ( 1.5 2.7 ( 0.8 1.2 ( 0.4
3.9 ( 0.2 8.7 ( 0.9 1.5 ( 0.1 0.069 ( 0.005
2.2 ( 0.2 8.5 ( 1.8 22 ( 4 33 ( 2
5.1 ( 0.2 4.7 ( 0.3 0.70 ( 0.05 0.044 ( 0.001
1a 1a 1a 1a 0.92 ( 0.06 0.87 ( 0.095 1.7 ( 0.4 1b
a Data were fit with RH fixed at 1 because 0 M NaCl data were well described by this simpler model (14). b Insufficient data were available to determine RH, so its value was fixed at 1.
Table 2: Effect of 1 M NaCl and Various Metal Ions on kobs at pH 7.0 kobs (min-1) salta Ca2+
10 mM 10 mM Mg2+ 10 mM Sr2+ 10 mM Ba2+ 10 mM Mn2+ 10 mM Co2+ 10 mM Ni2+ 10 mM Cu2+ 1 mM Zn2+ 1 mM Pb2+ 1 mM Co(NH3)63+ none
-NaCl
+1 M NaCl
7.0 ( 0.4 3.3 ( 0.2 1.2 ( 0.1 0.068 ( 0.002 0.77 ( 0.06 0.34 ( 0.03
